Data Center Knowledge | News and analysis for the data center industry - Industr's Journal
[Most Recent Entries]
[Calendar View]
Friday, May 16th, 2014
| Time |
Event |
| 12:00p |
HFN Lights Up Wireless Route Between Two Major New Jersey Trading Hubs Hudson Fiber Network has lit up a new wireless network connection between two of northern New Jersey’s most important high-frequency trading hubs: Equinix’s NY4 data center in Secaucus and CenturyLink Technology Services’ NJ2 facility in Weehawken.
The wireless route provides redundancy to the physical fiber route HFN has been operating between the two facilities, but the company says it is also about 40 percent faster than the fastest fiber route. That kind of reduction in latency has attracted interest from a multitude of financial-services companies, Keith Muller, HFN’s CEO and co-founder, said.
The NY4 data center at 755 Secaucus Road in Secaucus hosts primary infrastructure of exchanges operated by BATS Global Markets. The operator uses the facility to house BATS – the third-largest equity market in the U.S. – as well as EDGA and EDGX, which are securities exchanges BATS gained control of when it merged with their owner Direct Edge in August 2013.
HFN likely to benefit from BATS relocation to NY4
BATS was hosted at CenturyLink’s NJ2 data center at 300 Boulevard East in Weehawken before the merger, but in February of this year the company said it would move the exchange to NY4, where Direct Edge exchanges were located, to consolidate operations.
HFN may benefit from this relocation, since financial service companies with large footprints at NJ2 may want a low-latency connectivity option to NY4 to connect to the BATS exchanges without moving their servers, Muller said.
The route does not have too much competition, since the only other direct connections between the buildings are operated by two financial institutions, Muller said. He declined to say who they were, but BATS said it would keep a Point of Presence (PoP) at NJ2 to offer members direct connectivity to its exchanges at NY4. The other provider is most likely Nasdaq OMX, since it advertises connectivity services (including wireless connectivity) in both locations.
From Secaucus to Weehawken in 30.5 microseconds
HFN conducted the first latency tests on the new radio-frequency route about one week ago, reporting latency of 30.5 microseconds. That is latency before the signal enters the company’s Enyx box, which splits the service into different bandwidths, Muller said. The box adds another 4 microseconds, so total latency before the signal enters customer equipment is 34.5 microseconds, which is still a lot faster than the 59-microsecond latency on HFN’s fastest fiber route between the two locations.
The Enyx box also manages failover between fiber and wireless service in case there is an interruption to either one. “The unique thing about that box is when we have the fiber plugged into it, it’s an active-active-type solution,” Muller said. That means both the fiber and the hybrid route are constantly active, and packets lost on one route are picked up by the other and the data is reassembled at the other end.
HFN offers similar “hybrid” connectivity to the NASDAQ OMX data center in Carteret and to the NYSE Euronext facility in Mahwah.
The company uses wireless connectivity technology by ULL Networks – a Cary, North Carolina-based company. HFN is an exclusive provider of ULL services in the New York-New Jersey metro. When the two companies announced the partnership in April 2013, they said the goal was to eventually extend ULL services provided by HFN to four Equinix facilities in the metro: NY1 and NY4 in New Jersey and NY8 and NY9 in New York City.
NY8 and NY9 are located at 60 Hudson Street and 111 8th Avenue, respectively – New York’s most important network interconnection hubs. | | 12:30p |
Cisco Invests in Cloud Security Startup OpenDNS’ $35M Round OpenDNS, the predictive threat intelligence startup, has some dough to keep it going. The company has closed $35 million Series C round, from existing investors as well as new ones, the latter group including Cisco.
The existing investors are Greylock Partners, Sequoia Capital and Sutter Hill Ventures, and new investors on board besides Cisco are Glynn Capital, Evolution Equity, Lumia Capital, Mohr Davidow Ventures and Northgate Capital.
OpenDNS provides threat and policy management via DNS. Its portfolio is centered on two technologies: a global delivery mechanism for security called Umbrella and an advanced predictive threat intelligence platform called Security Graph. Combined, they aim to protect against a wide and changing landscape of Internet threats.
OpenDNS has 50 million daily active users and more than 10,000 enterprise customers. The company uses big-data analytics and machine learning to discover and predict when and where new attacks on the Internet are being stages.
“The erosion of the corporate IT perimeter brought on by cloud computing, mobile computing and the use of personal access devices demands a new type of network security that is always on, friendly to the employee, and doesn’t hamper performance, wherever the user may be,” said David Ulevitch, CEO of OpenDNS. “We identified this market opportunity several years ago.”
Cisco is closely tracking enterprise data and mobility security, which probably helped land OpenDNS on its radar. The startup’s roots are in the consumer space but it has been moving into the enterprise, which was also likely to be an enticing factor for the networking giant. OpenDNS is doing interesting things around security at the heart of Cisco’s playground: the network.
Cisco has a growing cloud security play. It launched Managed Threat Defense in April, a managed security solution that applies real-time, predictive analytics to detect attacks and protect against advanced malware across customers’ extended networks.
“OpenDNS is the future of enterprise security, protecting workers the way they work today,” said Stefan Dyckerhoff, managing director of Sutter Hill Ventures and former senior executive at both Cisco and its competitor Juniper Networks. “OpenDNS is forging new ground in enterprise threat protection with a cloud-delivered, predictive security service that protects all devices, anywhere, anytime without sacrificing performance or manageability.”
| | 1:00p |
Canonical and Ubuntu in the OpenStack Landscape Canonical, famous for its distribution of the Linux operating system called Ubuntu, has also been very active with the popular open source cloud operating system OpenStack. The company announced a number of things at the OpenStack Summit in Atlanta this week. Some announcements demonstrated the strength and speed of Ubuntu on OpenStack at hyperscale and some were aimed at making it easier to adopt OpenStack and improving interoperability.
“There’s a lot of talk interoperability, ecosystem and speed this year,” Mark Baker, Ubuntu Server and Cloud product manager, said. Ubuntu has been closely tied to OpenStack from the beginning, acting as reference OS throughout the cloud infrastructure software’s development. It is still a large part of the landscape: a survey at the close of the summit revealed that the Ubuntu platform was on more than half of OpenStack deployments.
More OpenStack clouds in production
The big difference between previous conferences and this year’s, according to Baker, was the makeup of attendees. “Previous events were biased toward vendors and technology companies. We were all sitting around collaborating, but there was a lower percentage of real users in the past. The balance has swung the other way now, where the users dominate.”
Not only is OpenStack drawing real attention from users, these users are also more and more frequently running OpenStack in production environments. “We’ve had a lot of customers building clouds [with] 20 to 50 nodes,” Baker said. “We have now seen in the last six months a huge rollout on the part of many, pushing it up to 200-400 nodes. There’s a level of comfort there that they can run it into production.”
Canonical has had a lot of uptake for Ubuntu on OpenStack. “We worked closely with the early adopters in the telecommunications space: Orange, NTT, ChinaTel, etc. We are strong in that space. We’re making good headway. We’re also seeing a push into financial services, like Wells Fargo, through integration with VMware. We’re more of a natural ally for VMware.”
That is because Canonical’s approach is one of working with other vendors, focusing on interoperability, complementary technologies and taking an overall ecosystem approach. This is an alternative to building a tight “enterprise stack,” where the sell is one vendor building many components, tightly integrated.
A mobile OpenStack sandbox
At the summit, the company introduced The Orange Box, which is a complete mobile Ubuntu OpenStack cluster with 10 nodes, 4 cores, 16GB of RAM and SSD storage, to deploy infrastructure on premises for testing or proof of concept. The goal is to get companies up and running with OpenStack easily.
“It can be tricky building an OpenStack cluster in an enterprise,” Baker said. “You need a bunch of machines, and there are different groups, different teams negotiating. It’s hard in an organization to get started. This box was designed to help them. It will bring them up with an OpenStack sandbox. It’s very much part of our jumpstart program.”
There is also a new jump-start program: a two-day Ubuntu OpenStack training course using Orange Boxes.”We’ve been offering training for a little while but the main addition is the Orange Box. We want to continue to drive the adoption, continue to make it easier for enterprises to get up to speed.”
Interoperability Improvements
Ubuntu’s Juju orchestration tool and MAAS (Metal as a Service, a provisioning construct created by Canonical) will now support Windows guest images and CentOS. “The key part is it can now orchestrate services on other other operating systems … and deploy services onto OpenStack,” Baker said.”Using the same tool becomes extremely powerful. Customers are wise to interoperability. When they choose cloud technology they’re making a choice that will have a big impact on the next five to 10 years.”
Canonical also demonstrated Ubuntu running on 64-bit ARM processors. The company is also working closely with IBM Power architectures, as well as other collaboration with IBM in general. “OpenStack is growing rapidly. Clearly Intel is the dominant processor of choice, but there’s a diverse community out there. Some are going to want to have choice of architectures.”
The company is also now offering fully managed OpenStack, with infrastructure, tools, software support and monitoring at $15 per host per day. “What we provide is the expertise and people to build, manage, scale and continue to manage it,” Baker said.
Flexing muscle
Canonical made a number of announcements that demonstrate high performance of infrastructure running its OpenStack distribution. The company said that the world’s fastest supercomputer, Tianhe-2 at the National University of Defense Technology in China, is running Ubuntu OpenStack, and announced a record benchmark for hyperscale computing.
AMD broke industry records running Ubuntu OpenStack on its SeaMicro servers, deploying 75,000 virtual machines in 6.5 hours, which it said was the fastest yet for OpenStack. | | 2:30p |
Data Center Jobs: TierPoint At the Data Center Jobs Board, we have a new job listing from TierPoint, which is seeking a Facilities Engineer in Seattle, Washington.
The Facilities Engineer is responsible for implementing customer sales orders and requests. This includes, but is not limited to installation of low voltage CAT6 structured wiring, coax, and optical fiber as well as DC and AC line voltage. This position is also responsible for the maintenance and testing of all Data Center electrical, cooling and general support infrastructure. Working with generators, uninterruptable power supplies, switchgear, electrical distribution and air conditioning units are normal daily functions. To view full details and apply, see job listing details.
Are you hiring for your data center? You can list your company’s job openings on the Data Center Jobs Board, and also track new openings via our jobs RSS feed. | | 2:40p |
Friday Funny: Caught Up in Cables Thank Goodness It’s Friday! That means there’s a weekend ahead, and there’s a Friday Funny. Time for some laughs.
Diane Alber, the Arizona artist who created Kip and Gary, has a new cartoon for Data Center Knowledge’s cartoon caption contest. We challenge our readers to submit a humorous and clever caption that fits the comedic situation. Please add your entry in the comments below. Then, next week, our readers will vote for the best submission.
Here’s Diane’s thoughts on this week’s cartoon, “I just love spaghetti cabling comics and sometimes when I’ve toured a data center where cable is EVERYWHERE. I have often thought there could be someone trapped and no one would know!”
Congrats to the last cartoon winner, Michael Gray, who won with, “I had to start a BitCoin farm to support my coffee addiction.”
For more cartoons on DCK, see our Humor Channel. For more of Diane’s work visit Kip and Gary‘s website. | | 4:30p |
VCE Adds Cisco’s ACI to Vblock Converged Infrastructure Converged infrastructure vendor VCE unveiled new Vblock systems with networking capabilities to support Cisco‘s Application Centric Infrastructure (ACI). Combined with VCE Vision Intelligent Operations software, the new systems can be built with Cisco Nexus 9000 series switches, which will be the first ACI-enabled switches.
Cisco announced ACI in 2013. It was the software defined networking product of Cisco’s “spin-in” Insieme Networks, which the networking giant acquired last year. ACI is radically different in approach to SDN than the one that dominates the SDN movement, which is centered around the open SDN protocol OpenFlow.
The new VCE Vblock Systems 340 and 720 with Cisco Nexus 9000-series switches are available for ordering immediately. VCE said they will be ACI-enabled upon general availability of Cisco APIC later this year.
Research from Wikibon shows that the converged systems market will reach $402 billion by 2017. VCE recently announced that it surpassed its 2013 goal of generating $1 billion in annual sales. VCE’s roots grow out of Cisco, which created the company together with EMC, VMware and Intel to sell Vblock. The product combines Cisco’s Unified Computing System servers and switches, EMC storage and VMware virtualization software.
Todd Pavone, VCE’s executive vice president for product strategy and development, said, “Combining the value ofVblock Systems with Cisco Application Centric Infrastructure, we can now deliver a holistic architecture with centralized automation and policy-driven application profiles, resulting in greater application predictability and availability with significantly lower costs.”
VCE Vblock systems with Cisco Application Policy Infrastructure Controller (APIC) and Nexus 9000 switches will automate provisioning of the complete application deployment lifecycle. These “ACI-ready” Vblock systems will take full advantage of policy-based templates that automate the provisioning of network resources, application services, security, segmentation and workload placement.
Vblock systems are individually optimized for a variety of workloads. Its modular architecture scales as workloads increase, giving a means to cost-efficiently extend compute, network or storage resources to increase performance and capacity. | | 5:04p |
Internap Data Center Outage Takes Down Livestream, StackExchange Internap experienced a power outage at its data center located in the Google-owned building at 111 8th Avenue in New York City. There appears to have been a utility power outage, followed by a problem with uninterrupted power supply system that supported the Internap facility. The outage affected about 20 customers, according to Internap.
The outage brought down a few notable Internap customers, including Livestream, a popular streaming video platform, for most of the morning Friday. It also affected the StackExchange network of sites, popular among developers. So far, the customers have been more vocal about the outage than Internap has been.
Livestream first reported issues at 3:49 a.m. EST. The company continued to report problems both on its status page and through live tweets throughout the morning. As of 11:30 EST the company reported that power was restored and systems were being brought back online.
“We have experienced a power outage with our data center provider Internap. We are still working to restore power as quickly as possible,” Livestream said in an update. The data center engineers are trying to restore power.
StackExchange reported issues on its status page as well as its Twitter feed. The company first failed over to its Oregon data center, but incurred further problems when migrating back to the main New York data center. “We are in a holding pattern since the UPS system serving both our primary and backup power feeds (yes, we’re very excited to learn this too) is still being worked on,” Stack Exchange Network said.
After the hard power outage, the problem appeared to be with the UPS system, something that has affected Internap in the past. Utility power to the building has been restored, but Internap said engineers were still working with its electrical vendors to stabilize UPS power.
In 2009, Internap suffered an outage in Boston that resulted from poor battery maintenance. The outage caused finger pointing between Internap and VoIP provider Ooma that resulted in some bad publicity.
How Internap handles this morning’s situation will be integral in minimizing the negative impact of the outage. Livestream is a sizeable player and can raise a ruckus if the situation is not handled correctly.
Outages are never a good thing, but transparency, frequent communication and taking responsibility goes a long way in retaining customers. If the provider does not come clean about causes of an outage, chances are a customer will be more vocal about it.
In addition to 111 8th Avenue, Internap’s New York metro data centers include a facility at 75 Broad Street in New York and another one across the Hudson River at 1 Enterprise Avenue North in Secaucus, New Jersey.
UPDATE: Internap issued the following statement: On Friday, May 16 at approximately 3 a.m. ET, Internap experienced a power outage in a portion of its data center located at 111 8th Avenue in New York City. Utility power was restored at approximately 6 a.m. ET. The outage impacted approximately 20 customers using a combination of colocation and IP connectivity services. We are still working to determine the root cause of the disruption, but the site has been returned to normal N+1 redundant configuration. We have been providing customers with hourly status notifications since the beginning of the event, and our technicians are onsite 24×7 to assist them. Customers can contact noc@internap.com with questions. |
|